Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
' ~- - 1 - 20~'~6~2
Background of the Invention
Summary of the Invention
An improved spiral wound membrane element comprising layers
of membrane material fluid tight sealed along 3 edges enclosing a
permeate spacer creating at least one permeate envelope upon which a
feed/retentate spacer layer is laid, along at least one membrane face,
the entire multi layer arrangement being wound around a hollow central
. mandrel (which may be closed at one end) and to which the permeate
envelope is in fluid communication through the fourth unsealed edge,
creating a spiral wound module element is disclosed which is useful in
separation processes wherein a pressure gradient is maintained across
the membrane from a feed side to a permeate side, the improvement
comprising using as the feed/retentate spacer at least one layer of a
material having an open cross-sectional area of at least 30-70fo and
using as the permeate spacer at least three layers of material charac-
teri~ed in that the outer layers are a fine material having an open
cross-sectional area of about 10 to 50% and a coarse layer having an
open cross-sectional area of about 50 to 90% interposed between the
aforesaid fine outer layers and wherein the fine outer layers are in
interface contact with the membrane layers enclosing the permeate
spacer.
Description of the Related Art
Spiral wound elements contain permeate and retentate spacers
as a routine matter of standard element design, see, e.g. USP
3,417,870. Various attempts have been made to improve the spacer
materials. Thus, USP 4,861,487 describes a low pressure drop spacer
composed of generally parallel elongated filaments positioned general-
ly parallel to the flow direction of the feed stream and wherein the
elongated filaments are connected by shorter bridge filaments which
are placed at an angle to the flow of the feed stream to provide for a
low pressure drop. European Patent Application 89305966.7 (publica-
tion number 347174) describes a spiral wound membrane cartridge
wherein the feed spacer material having a plurality of parallel ribs
- 2 - 2a9'~632
extending in an axial direction, interconnected by a matrix of smaller
filaments generally perpendicular to the parallel ribs which results
in a reduction to the flow resistance. W091/11249 describes a spiral
wound element which utilizes a divided central mandrel and a permeate
region which employs a high density porous spacer flanked on two sides
with low density porous spacers.
US 5,069,793 describes a spiral wound element for use in
pervaporation designed to produce maximum permeate flow throughput per
element volume. This is achieved by use of a permeate spacer selected
to take advantage of the fact that the total permeate flow throughput
from a module passes through a maximum as the resistance to vapor
transport of the permeate spacer material is progressively decreased.
The capability of the permeate spacer material to transport permeating
vapor from the membrane surface to the permeate collection pipe is
expressed as a normalized conductivity, or permeate vapor flow, per
unit pressure drop in the permeate channel, per unit transmembrane
flux. The permeate channel is defined to use a spacer material such
that permeate flow throughput is 60-90% of the maximum possible value.
The permeate spacer can be a sheet of material having a cross-section-
al thickener which varies, giving from relatively thin at the far edge
to thick at the edge adjacent the central mandrel. Alternatively the
spacer can be made of multiple layers of the same or different spacer
material.
Description of the Invention
In a spiral wound membrane separation element comprising a
hollow central mandrel (which may be closed at one end) around which
are wound multiple layers of membrane, feed spacers and permeate
spacers wherein layers of membrane surround a permeate spacer, said
membrane layers being fluid tight sealed along 3 edges producing a
permeate envelope leaf, wherein multiple permeate envelope leaves are
attached along their fourth unsealed edge in fluid communication with
the interior of the hollow central mandrel, and a layer of feed/-
retentate spacer material extends along the outer surfaces of each
permeate envelope leaf, the spiral winding of multiple permeate
- 3 - 2a~'~~32
envelope leaves/feed-retentate spacers being wrapped with an outer
wrap layer to prevent unwinding and the end of the winding being
capped by an anti-telescoping device attached at the downstream end to
prevent telescopic displacement of the spiral wound layers during use,
the improvement comprising using as the feed/retentate spacer material
at least one layer of a material having an open cross-sectional area
of at least about 30 to 70fo, preferably 30 to 50%, preferably using
two layers of material placed between adjacent permeate envelope
leaves, which feed/retentate spacer material can be of either the same
or different material and of the same or different cross-sectional
area, preferably such feed/retentate spacer material being insulated
from the membrane surface by an interposed layer of chemically and
thermally inert woven or non-woven fabric about 1 to 15 mils thick
having a weight of about 0.5 to 10 oz/sq. yard and a Frazier air
permeability in the range 0.5 to 1000 cfm/sq. foot at 1/2 inch water
pressure, (example of a non-woven material being Nomex), such that
upon winding the multiple permeate envelope leaves and interposed
multiple feed/retentate spacer layer one obtains two layer of feed/-
retentate spacer material between adjacent permeate envelope leaves in
the winding, and using as the permeate spacer three or more layers of
material, the outer layers which are in contact with the membrane
(i.e. with the membrane surface per se or with the integral backing of
the membrane if the membrane is cast on a backing, this backing not
being counted as one of the spacer layers) being a fine material
having an open cross-sectional area of at least about 10 to 50%,
preferably at least about 10 to 30% and interposed between the fine
outer layers will be coarse layers having an open cross-sectional area
of at least about 50 to 90%, preferably about 60 to 90%. The multi-
layer permeate spacer comprises at least 3 layers but may comprise 3
to up to 7 layers alternating fine and coarse provided that the outer
layers in contact with the membrane are fine material layers. It is
preferred that an odd number of layers be used to minimize intermesh-
ing but an even number of layers can also be employed in which case it
is preferred that the layers be of materials of different mesh size so
as to prevent or minimize intermeshing of the layers. The limit on
the number of layers used in fabricating the permeate spacer layer in
each permeate envelope leaf, the thickness of each leaf, the length of
each leaf, the number of leaves attached to the central mandrel and
the number and thickness of the feed/retentate spacer between adjacent
permeate envelope leaves will be set as a compromise among competing
factors including the ability to ultimately wind the assembly around
the central mandrel, the pressure drop along the length of each
permeate envelope leaf as well as across the feed spacer and the
membrane surface area obtainable in each spiral wound module.
If too many layers are used or if the layers used are too
thick it will become difficult to wind the spiral wound element.
Also, obviously too thick a permeate envelope will negatively impact
on the total membrane surface area available in the final spiral wound
element of a given diameter.
The preferred number of permeate spacer layers is 3 to 5.
The multi-layer permeate spacer may be sized slightly
smaller in its dimensions than the membrane layer surrounding it so
that the spacer does not intrude into the area between the membranes
at the three edges along which the membrane layers are fluid tight
sealed. Intrusion of permeate spacer into this area interferes with
the ability to effectively seal the membrane edges to create the
permeate envelope.
The preferred number of feed/retentate spacer layers is 2.
In preparing the element it has been found that interposing a layer of
chemically and thermally stable woven or non-woven material about 1 to
15 mils thick weighing about 0.5 to 10 oz/sq. yard and having a
Frazier air permeability in the range 0.5 to 1000 cfm/sq. ft at 1/2
inch water pressure interposed between the feed/retentate spacer
material and the surface of the membrane and the permeate spacer
material and the surface of the membrane improves membrane element
long term performance and improves the vacuum tightness of the
resulting spiral wound package. This interposing layer acts as a
shield between the membrane surface and the feed/retentate spacer
layer and/or permeate spacer. For low temperature applications
polyethylene, polypropylene, nylon, etc. felt can be used as the
~.... _ 5 _
shield. For high temperature applications the choice is more limited, with
polyamide (e.g. Nomex which is a blend of high temperature nylon and
polyester), teflon, fiberglass or mixtures thereof being suitable candidates.
This/these shield layers are not included in the count of permeate
spacer layer or feed/retentate spacer layers. When used on the feed side the
shield layer protects the membrane from being punctured by the feed spacer;
when used on the permeate side it protects the membrane from the support
mesh/permeate spacer layers.
The layer of feed/retentate material extended on the surface of at
least one face of the permeate envelope is substantially equivalent in its
dimensions in term of length and width to the permeate envelope.
Adhesives are used in preparing the spiral wound element. Different
adhesives for different types of applications and environments have been
identified and are described in USP 4,464,494 and USP 4,582,726 incorporated
herein by reference. Various other adhesives such as high temperature epoxy
(e.g. Tra-bond 2125 from Tra-Con or Duralco 4400, 4525, 4700, 4703 from
Cortronics Corp.) or non-epoxy adhesives (e.g. alumina/zirconia/ceramic
adhesives such as Resbond 903 HP, 904 Zirconia, 904 Quartz and 906
Magnesia from Cortronics Corp.) may also be used.
The spiral wound module wrapped in its outer wrap and fitted with the
anti-telescoping device can be inserted into a pressure vessel having an
internal
diameter equal to the exterior diameter of the module, and long enough to hold
from one to any number of modules in series, said pressure vessel being fitted
with feed entrance/retentate exit means and separate manifold means for
recovering permeate from the open end of the hollow central mandrel.
Alternatively, multiple modules can be installed in parallel within a single
containment vessel.
_6-
20~'~632
In producing the spiral wrapped modules of the present
invention having feed/retentate and permeate spacers as described
various materials of construction can be used to meet the required
spacer characteristics.
The feed/retentate spacer can be a woven mesh material or a
non-woven mesh material, e.g., a first layer of parallel spaced apart
filaments covered by a second layer of parallel spaced apart filaments
laying perpendicular or diagonally to the first layer wherein the
filaments of the first and second layer are attached to each other at
their points of contact, such a material hereinafter referred to as
non-interwoven filament material.
When using such mesh materials as the feed/retentate spacer
the spacer will comprise a single layer or multiple layers of material
at least one of which is 16 to 80 mesh, preferably 16 to 60 mesh, more
preferably 20 to 60 mesh and between 10 to 30 mils thick preferably
17-25 mils thick. For ease of fabrication it is preferred that 2
layers of material be used as the feed/retentate spacer both layers
being preferably made of the same material. Use can be made of
different mesh sizes to prevent intermeshing. If more than 3 layers
are used, the layer in contact with the faces of the membranes would
be a finer material and the layer between these face contacting layers
would be a coarser material within the aforesaid limits, e.g., the
fine layers could have a 50 to 80 mesh while the coarser layer could
have a 20 to 50 mesh. The materials) used will be such as to provide
a feed spacer having an open cross-sectional area of at least 30 to
70%, preferably about 30 to 500. As previously stated, it is pre-
ferred that the membrane and the feed/retentate spacer be separated
from direct contact by an insulating layer of chemically and thermally
inert woven or non-woven fabric such as Nomex.
A problem encountered when using multiple elements in series
is that the feed/retentate flow rate through the end elements is low
since a significant portion of the feed would have permeated across
the membrane in the first few elements. This results in low feed
velocity through the end elements and the performance of these
' 2~~~~3~
elements is compromised. The low velocity through the end elements is
aggravated when a relatively high open area aluminum (30 mesh-0.01"
wire diameter - 49% open area) screen is used as the feed spacer in
the spiral wound element design. This screen gives low feed-to-
retentate pressure drop, which is an important consideration with six
elements in series. With this element design, the overall pressure
drop across the elements and with two intermediate heat exchangers
used to reheat the feed is expected to be less than 15 psi.
The feed velocity through the elements can be increased by
using a lower open area screen as the feed spacer. An example of such
a material would be 50 mesh - 0.0090" wire diameter stainless steel
screen which has an open area of 300. When two layers of this screen
are used as the feed spacer, the feed-to-retentate pressure drop is
4.6 psi at 10 kg/min feed rate. This represents a significant
increase in pressure drop versus the previously identified design with
a single layer of 30 mesh aluminum feed spacer. Although the 2 x 50
mesh screen design is excellent for increasing the feed velocity thus
creating turbulence, a disadvantage is that the overall pressure drop
when using multiple elements, e.g., with six elements in series and
two intermediate heat exchangers, is well over 40 psi if this element
design were used. Since it is necessary to maintain at least 10 psi
pressure on the retentate, it would then be necessary to operate the
lead element at over 50 psi inlet pressure. This is not acceptable
since this would exceed the maximum tolerable pressure of pervapora-
tion spiral wound elements which is around 40 psi.
In a preferred pervaporation process the elements are staged
by using increasingly higher pressure drop feed spacers in order to
get high feed velocity through the end elements. With this staged
pervaporation process, a relatively high open area feed spacer such as
30 mesh aluminum would be used for the first four elements while a
relatively low open area feed spacer such as 2 x 50 mesh stainless
steel would be used for the last two elements. With this combination,
the overall pressure drop for the system is expected to be less than
25 psi, which would be acceptable. More importantly it can also be
~.. _ g _
expected that the performance of the end elements to be greatly
improved since the feed velocity would be high.
Another example would be to use 30 mesh aluminum as the feed
spacer for the first two elements, 40 mesh aluminum feed spacer for
the second two elements, and 2 x 50 mesh stainless steel feed spacer
for the last two elements. Needless to say, there are numerous other
ways to stage the elements with increasingly higher pressure drop feed
spacers in order to achieve the desired high velocity through the end
elements.
This use of increasing pressure drop feed/retentate spacers
in the down stream elements of multiple spiral wound elements in
series should also be useful for reverse osmosis and ultrafiltration
wherein the performance is especially sensitive to feed velocity.
The permeate spacer material used can also be selected from
the aforesaid woven or non-interwoven filament materials. As previ-
ously stated the spacer comprises an assembly of three or more layers,
alternating fine and coarse material. The fine material which
supports the membrane and prevents intrusion into the permeate spacer
can be a woven or non-interwoven filament material having at least a
50 mesh or finer, preferably 60 to 300 mesh, more preferably 60 to 150
mesh, still more preferably 80 to 120 mesh, most preferably 100 to 120
mesh and about 3 to 15 mils thick. The coarse material can also be a
woven or non-interwoven filament material having less than a 80 mesh,
preferably less than 50 mesh, more preferably less than 35 mesh, most
preferably less than 20 mesh, and from 10 to 30 mils thick preferably
from 17-25 mils thick, it being understood that in practice the fine
material used will have a finer mesh than the coarse material used.
Likewise when using fine mesh material in the 200-300 mesh range it is
preferred that the coarse layer be in the 30 to 80 mesh range when the
element is to be used at elevated temperature and pressure.
The permeate spacer can comprise 3 or more layers. When 3
layers are employed two layers of fine material (support layers) are
used in contact with the membrane layer and a coarse layer is
- ~0976~2
interposed between the two fine layers. If 4 layers of spacer
material are used, the two outer layers in contact with the membrane
are still the fine material support layer and coarse material consti-
tute the two inner layers interposed between the two fine outer
layers. Care should be taken when using this 4 layer embodiment to
insure that the two coarse layers which are in contact with each other
either have different cross-sectional profiles or of the same cross-
sectional profile are out of register one with the other to insure
that the coarse materials do not intermesh with each other which if
that happened would result in a substantial reduction in the open
cross-sectional area of the materials available of permeate flow. In
such an intermesh situation flow would be inhibited and an undesired
pressure drop across the permeate spacer resulting in reduced flux
would be encountered. If five layers are employed they would be
arranged in a fine/coarse/fine/coarse/fine sequence wherein the three
fine layers could be the same or different materials of the same or
different fine cross-sectional area, within the previous definition of
fine material; likewise the two coarse layers could be the same or
different materials of the same or different coarse cross-sectional
area, again within the previous definition of coarse material.
The spacer materials can be made from any plastic or metal,
e.g., polyester, polysulfone, polyester, nylon, teflon, etc., or fiber
glass, or stainless steel, aluminum or brass etc. In general any
material which will be chemically inert and thermally stable in the
intended environment of use of the final element can be employed as a
material of construction. It is preferred however that the spacer
material be made of metal, e.g. aluminum or stainless steel and more
preferably that it be steel especially in the case of the fine mesh
material. In order to insure optimum operability of the final element
the spacer should be capable of preventing membrane intrusion into the
permeate space under the pressures employed. This ability to prevent
membrane intrusion has been correlated to spacer stiffness. An
available measure of stiffness is the tensile modulus of elasticity.
The stiffness of a number of common spacer materials is presented
below:
- to -
Polyester 2-3 x 105 lbs/sq. inch
Aluminum 10 x 106 lbs/sq. inch
Stainless steel 28 x 106 lbs/sq. inch
Thus, in the present invention the fine permeate spacer
support material, in addition to having the recited open cross
sectional area, has a stiffness of at least about 2-3 x 105 lbs/sq.
inch, preferably at least about 10 x 106 lbs/sq. inch, most preferably
at least about 28 x 106 lbs/sq. inch and higher, the fine permeate
spacer support material being most preferably stainless steel.
The stiffer material provides better support which elimi-
nates or minimizes intrusion which in turn minimizes permeate pressure
drop. This is especially true at operating conditions (i.e. in
permeate at 140°+C). This is to be compared to the performance of a
less stiff material such as polyester. Because of the low stiffness
factor of polyester, both the membrane and the polyester support are
pushed into the permeate spacer channels, especially at higher temper-
atures and/or pressures. Thus if one considers material stiffness
with mesh size, a finer material of greater stiffness can be used with
a more coarse layer than can a similar fine mesh material of lesser
stiffness.
For example while a 200 mesh aluminum support may work
satisfactorily with an 80 mesh coarse layer, the 200 mesh aluminum
support would not be satisfactory with a 17-50 mesh coarse layer.
However, a 200 mesh stainless steel support would be satisfactory with
a 30-80 mesh layer because of its greater stiffness. Specific selec-
tions of materials within the aforesaid recitations are left to the
practitioner to make with consideration being paid to the temperature
and pressure of element application and the design or target permeate
pressure drop across the element.
When the element is to be used for pervaporation it is
preferred that the fine material used as permeate spacer support be in
the 60 to 150 mesh, preferably 80 to 120 mesh range 5 to 15 mils thick
- 11 -
and be of stainless steel while the coarse material has a mesh size of
less than 50 mesh and is 15 to 30 mils thick.
In addition to the woven or non-interwoven filament mate-
rials previously described, the spacer materials having the necessary
cross-sectional areas can be materials which exhibit no mesh but
rather are spaced apart ribs running in parallel on a thin solid
support sheet. Such sheets can be fabricated by casting or extruding
with the aforesaid ribs cast or extruded as integral parts of the
sheet. Alternatively individual filaments can be deposited on a
pre-existing sheet. The channels defined by the spaces between the
parallel ribs or filaments and the height of the ribs or filaments
would provide the cross-sectional areas falling within the aforesaid
definitions. Use of such materials would require that the sheets be
oriented in the permeate envelope such that the channels would be
aligned in the direction of permeate flow in the envelope into the
hollow central mandrel.
By the practice of the present invention performance of the
spiral wound element in terms of both flux and selectivity is nearly
identical to that of the membrane when used by itself, uninfluenced by
any hydrodynamic effects introduced by spacer materials.
The present invention is especially useful in the separation
of aromatics from non-aromatics, such as in heavy cat naphtha separa-
tion, intermediate cat naphtha separation, light cat naphtha separa-
tion etc.
Membranes which are useful in such separations include
polyurea urethane disclosed and claimed in U.S. Patent 4,914,064,
polyurethane imides disclosed and claimed in U.S. Patent 4,929,358,
polyester imides disclosed and claimed in U.S. Patent 4,944,880,
isocyanurate crosslinked polyurethane membranes, disclosed and claimed
in U.S. Patent 4,983,338 and U.S. Patent 4,929,357, polyester mem-
branes disclosed and claimed in U.S. Patent 4,976,868, preferably the
polyester imides of U.S. Patent 4,944,880 and U.S. Patent 4,990,275,
all of which are incorporated herein by reference. Polyacrylate
m_ - I2 - _ 2~J'~~3~
membranes may also be used. Acrylic acid ester homopolymers or their
copolymers with each other or with acrylic acid can be formed into
membranes. The acrylic acid monomer units can be in free-acid form or
partly or totally neutralized with metal or alkylammonium ions. The
membranes can be covalently or sonically crosslinked.
It has been found that membranes, such as the polyester
imide membranes which when used in aromatics/non-aromatics separation
processes such as heavy cat naphtha separation lose performance
overtime due to the build up of a corrosion deposit layer (e. g. iron
sulfide) on the membrane can be restored to their original performance
levels by soaking the membrane in a gasoline dispersant/detergent,
which is a surface active material having a molecular weight in the
range from 500 to 3000. The dispersant/detergent has a backbone which
can be polybutene or polypropylene, bearing with functional groups
comprising ether amines, hydrocarbonyl amines, hydrocarbonyl amides or
mixtures thereof. As example of a useable membrane regeneration
detergent/dispersant is CS-3 Kerofluid available from BASF.
It is also important in membrane separation processes
especially pervaporation processes that the membranes be defect free.
The presence of holes in membranes can significantly decrease membrane
selectivity performance. It has been found that micro defects in
pervaporation membranes can be identified before module or element
assembly by brushing the surface of the membrane with a liquid such as
heptane and pulling a vacuum or just brushing the surface of the
membrane with a water, isopropyl alcohol (IPA) mixture (e.g. 50/50 by
weight). Heptane or IPA passing through the defects wet the backing
of the membrane upon which the membrane is cast resulting in translu-
cent spots thus identifying the defect which can be patched by apply-
ing glue over the defect area.
The present invention is illustrated in the following
non-limiting Examples.
- 13 -
Example 1
An element wherein the feed/retentate and permeate spacers
consisted of 14 mesh polyester (the permeate spacer being separated
from the membrane by layers of Tricot 8846 polyester serving as
membrane support) was compared with an improved element within the
scope of the present invention wherein the 14 mesh polyester permeate
spacer was sandwiched between layers of 80 mesh stainless steel as
support, (no Tricot 8846 polyester support layer being employed) and
wherein the feed/retentate spacer was a 33 mesh Teflon. Tricot 8846
is available from Hornwood Inc. of Maryland. It is a woven fabric 4
mils thick having wales of 48 strands per inch and courses of 58
stands per inch. This fabric is coated with epoxy having a resin
pick-up of 16f°.
The two elements were evaluated for the separation of heavy
cat naphtha at 100°C and 10 mbars permeate pressure using a polyurea/-
urethane membrane.
The membrane was prepared as follows:
A solution containing a polyurea-urethane polymer is pre-
pared. Four point five six (4.56) grams (0.00228 moles) of polyethyl-
ene adipate (MW - 2000), 2.66 grams (0.00532 moles) of 500 MW polyeth-
ylene adipate and 3.81 grams (0.0152 moles) of 4,4'diphenylmethane
diisocyanate are added to a 250 ml flask equipped with a stirrer and
drying tube. The temperature is raised to 90°C and held for 2 hours
with stirring to produce an isocyanate-end capped prepolymer. Twenty
grams of dimethylformamide is added to this prepolymer and the mixture
is stirred until clear. One point five grams (0.0076 moles) of
4,4'diamino-diphenylmethane is dissolved in ten grams of dimethyl-
formamide and then added as a chain extender to the prepolymer solu-
tion. This mixture was then allowed to react at room temperature
(approx. 22°C) for 20 minutes. This solution was diluted to 5 wt%
such that the solution contained a 60/40 wt% blend of dimethylforma-
mide/acetone. The solution was allowed to stand for one week. The
'~~- - 14 -
2~~'~~32
viscosity of the aged solution was approximately 35 cps. After this
period of time one wt% Zonyl FSN (Dupont) fluorosurfactant was added
to the aged solution. A microporous teflon membrane (K-150 from Gore)
with nominal 0.1 micron pores, 75% porosity cast on a non-woven
Nomex/polyethylene terephthalate backing, the combination being 4 mils
thick was coated with the polymer solution in a continuous operation.
The coating was dried in an oven heated to 60'C. This technique
produced a composite membrane with a polyurea/urethane layer between 3
to 4 microns in thickness.
The results of the evaluation are presented in Table 1.
Table 1
PERFORMANCE DIFFERENCE BETWEEN
PER11APORATION ELEMENT DESIGNS
Spiral Wound Element Desi4n
Permeate Spacer support/ Yes/Tricot 8846 Yes/80 mesh
Mesh Size Stainless Steel
Permeate Spacer Mesh Size 14 14
Feed Spacer Mesh Size 14 33
Element Performance (1)
Selectivity, Delta RON 8.1 10.2
Flux, Kg/M2-Day 31 48
(1) At 100°C and 10 mbars permeate pressure.
As can be seen the table, with the spiral wound design using
polyester tricot spacer support, a selectivity of 8.1 was achieved.
In contrast, with the spiral wound design using stainless steel spacer
support, a selectivity of 10.2 was achieved. In aromatics/saturates
separation, selectivity is measured by the octane difference between
the permeate and the feed. The flux also improved significantly. It
can be expected that the performance difference between the packages
will be even greater at higher temperatures since flux would be
significantly higher.
- 15 - 2a~'~~32
After the evaluation of element performance using heavy cat
naphtha was completed, studies directly measuring the flow character-
istics of the permeate spacers were made. For these studies the outer
wrap layer around the element was removed and the permeate envelope of
one leaf was carefully opened to the atmosphere along the sealed edge
furthest removed from and parallel with the hollow central mandrel.
Various vacuum levels were then drawn on the hollow central mandrel
and the resulting air flow rates through the permeate spacer were
measured. The data are presented in Table 1B.
Table 1B
Comparison of Flow Characteristics of
Permeate Spacer Designs
vacuum level measured flow rate,1/min (air at STP)
in
central mandrel --------------------------------------
(i.e. pressure element element
drop in permeatewith polyester with stainless
spacer), mm Hg tricot support layersteel support layer
4.0 4.0
7.6 6.8
14.7 11.3
16.2 25.1
27.9 11.9
32.1 36.8
46.0 46.4
50.8 30.3
54.0 19.3
76.2 38.5
127. 53.5
140. 62.3
The data show that for a given pressure drop, the air flow through the
element with the stainless steel support layer is significantly
greater than for the element with the polyester Tricot 8846 support
layer. A spacer flow resistance parameter can be calculated from a
linear regression of the data. For the element, employing the
polyester Tricot layer the parameter is 2.30 mrn Hg/(1/min of STP air)
while for the element employing the stainless steel layer it is 0.96
mm Hg/(1/min of STP air). A low value of the flow resistance parame-
ter is desirable since it indicates reduced permeate pressure drop and
~.. - 16 - 2~~'~G~2
thus that lower average permeate absolute pressure exists. For
pervaporation, lower permeate absolute pressure is associated with
higher selectivity and higher flux.
Example 2
A separate example was performed to determine the effect of
feed/retentate spacer open cross-sectional area and thickness on
permeator performance.
Four elements were fabricated. Each element used the same
polyester-imide membrane material, which was made by first endcapping
one part of 2000 molecular weight polyethylene adipate (PEA) with two
parts of pyrometallic dianhydride (PMDA) and then reacting one part of
the endcapped polymer with methylene dianiline (MDA) to form a
polyamic acid. The polyamic acid was then coated onto a 0.1 micron
teflon sheet as previously described in Example 1. The viscosity of
the polyamic acid was in the range of 90-150 cps at room temperature.
After the polyamic acid was deposited onto the teflon sheet, the
polyamic acid was cured at 260°C for 7.25 minutes. The permeate
spacer used consisted of five layers of screens, 120 mesh stainless
steel/17 mesh aluminum/120 mesh stainless steel/17 mesh aluminum/120
mesh stainless steel.
The four elements were evaluated on heavy cat naphtha at
140°C and 10 mbar permeate pressure.
Feed spacer of various open cross-sectional area (mesh) and
thickness were used. The results are presented in Table 2.
- 17 - 2~~~~7~~
Table 2
EFFECT OF FEED SPACER ON
PERVAPORATION ELEMENT PERFORMANCE
Feed Spacer
Screen Size, Mesh 67 33 18 14
Thickness, Mils 10 21 17 33
Element Selectivity
Delta RON 8.9 10.2 11.8 8.1
It is seen that with thick/coarse spacer (33 mils/14 mesh)
the element performance is low because feed velocity is relatively low
at a given flow rate. Conversely with a thin/fine spacer (10 mils/67
mesh) element because of high pressure drop the performance, while
improved over that achieved with the thick/coarse spacer is not as
high as obtained with a spacer of more moderate thickness/and a
coarseness between fine and coarse.
The feed spacer in permeation elements, therefore, advanta-
geously ranges from 16 to 80 mesh, preferably 16 to 60 mesh, more
preferably 20 to 60 mesh and from 10 to 30 mils thick, preferably
17-25 mils thick.
Example 3
To further illustrate the effectiveness of the improved
element design, several elements were fabricated and tested for the
pervaporative separation of heavy cat naphtha. All test elements
employed the same membrane material, the polyester imide of Example 2
cast on the same teflon membrane backing as previously described.
Four sample elements were fabricated, A, B, C and D.
Element A employed single layers of 14 mesh polyester as
both the feed spacer and permeate spacer. When used as the permeate
spacer it was isolated from the membrane surfaces by two layers of
polyester felt spacer support (Tricot 8846), one layer on each side of
209702
- is -
the permeate spacer, to prevent damage to the membrane layers by the
coarse permeate spacer during element fabrication.
Elements B, C and D used multiple layers of permeate spacer
material which were isolated from the membrane surfaces by intervening
layers of Nomex, and either single or multiple layers of feed spacer
material between adjacent permeate envelope leaves, in all cases the
feed spacer material being a finer material having a mesh greater than
20.
Table 3 presents the details of the feed and permeate
spacers used in each of the four membrane elements and indicated the
selectivity of each element in terms of delta MON (motor octane
number) of the resulting permeate.
The elements were all tested in a recirculating pilot plant
with heavy cat naphtha at 140°C, 15 psi feed pressure and 10 mbars
permeate pressure. The effectiveness of each element was assessed by
measuring the difference in the motor octane number (delta MON)
between the permeate and the feed. Flux was not compared because the
HCN used in the tests had been exposed to oxygen which negatively
effects the flux performance of the membranes.
When used by itself in a flat circular test cell without any
feed or permeate spacers the membrane supported by a fine sintered
porous metal support, under the same conditions exhibited a selecti-
vity in terms of AMON of about 12.9.
As can be seen from Table 3, Element A exhibited a selecti-
vity in terms of AMON of 9.9, which is 3.0 MON lower than the membrane
by itself.
In comparison preferred elements B, C and D of the present
invention using stiffer spacer material and finer/stiffer feed/-
retentate spacer material exhibited selectivity in terms of aMON
ranging from 11.7 to 12.4, in all cases an element efficiency of over
90%.
- 19 - 20~~'~~32
It is expected that if the membrane was placed directly on
the 14 mesh polyester permeate spacer without any fine mesh support
intervening layer between the membrane and the 14 mesh polyester
spacer, the membrane would fully embed into the permeate spacer,
resulting in severely degraded performance, or be punctured by the
spacer, resulting in inoperability.
Table 3
PERFORMANCEOF PERVAPORATION
ELEMENT
ELEMENT FE ED SPACERPERMEATE SPACER SELECTIVITY,
N0. AR RANGEMENTARRANGEMENT DELTA MON
A 14 MESH PE PE FELT(Tricot 8846) 9.9
14 MESH PE
PE FELT
Nomex
B 33 MESH AL 120 MESH SS 11.7
33 MESH AL 17 MESH AL
17 MESH AL
120 MESH SS
Nomex
Nomex
C 33 MESH AL 100 MESH SS 11.7
17 MESH AL
100 MESH SS
17 MESH AL
100 MESH SS
Nomex
Nomex
D 50 MESH SS 100 MESH SS 12.4
50 MESH SS 17 MESH AL
100 MESH SS
17 MESH AL
100 MESH SS
Nomex
(1) Delta MON Selectivity of PEI Membrane alone is 12.9
(2) PE: Polyester
- 20 - -
Example 4
A number of spiral wound element packages were prepared to
evaluate the effect of putting a non-woven shield layer between the
membrane and the feed/retentate spacer layers. Each element used the
same membrane as described in Example 2. The results are presented
below:
Element I II
Nomex
Feed Spacer Arrangement 50 mesh SS 50 mesh SS
50 mesh SS 50 mesh SS
Nomex
Permeate Spacer Nomex Nomex
Arrangement 100 mesh SS 100 mesh SS
17 mesh AL 17 mesh AL
100 mesh SS 100 mesh SS
17 mesh AL 17 mesh AL
100 mesh SS 100 mesh SS
Nomex Nomex
Vacuum Drop Time (min) 4 21
The vacuum drop time is a measure of the tightness of the
spiral wound element package. In this test a 29" Hg vacuum is pulled
on the element. The vacuum pump is then turned off. The pressure
inside the element rises. The vacuum drop time is the time which it
took the element to go from 29" to 22" Hg vacuum. The longer the
vacuum drop time, the tighter is the element.
Elements II and III were evaluated for vacuum drop time and also for
the pervaporative separation of heavy cat naphtha. The pervaporation
test was conducted on heavy cat naphtha initially at 140°C for 2 to 10
days and then at 150°C for 20 to 21 days and 10 mn Hg vacuum pressure
at a flow rate of 1300 lbs/hour. The results from the 150°C runs are
reported in detail below.
"~' - 21 -
Element III II
Nomex
Feed 30 mesh AL 50 mesh SS
Spacer
Arrangement
50 mesh SS
Nomex
Permeate Nomex Nomex
Spacer
Arrangement 100 mesh SS 100 mesh SS
17 mesh AL 17 mesh AL
100 mesh SS 100 mesh SS
17 mesh AL 17 mesh AL
100 mesh SS 100 mesh SS
Nomex Nomex
Vacuum 4 21
Drop
Time
(min.)
Days 20 21
on
Oil
at
150C
Initial 288 229
Flux
(kg/m2/day)
Flux p per day (%) -0.27* 0.24*
dro
InitialPermeate RONC 100.5 100.4
Permeate
RONC
Drop
per -0.087 -0.018*
day
(%)
* Not
Statistically
Significant
(RONC Number r)
- Research Clea
Octane
As can be seen from the table above, at 150°C high tempera-
ture operation element III with no Nomex showed higher initial perfor-
mance, higher flux at constant selectivity as compared to element II
which has Nomex on the feed side. However, element III showed signif-
icant loss in selectivity with days on oil whereas element II showed
no statistically significant loss in selectivity with days on oil.
The flux stabilities of both elements were satisfactory.
In the above elements the 50 mesh SS is 11 mils thick, the
100 mesh SS is 9 mils thick, the 17 mesh AL is 23 mils thick, the 30
mesh AL is 24 mils thick. The Nomex layer is a non woven fabric from
Veratek Inc. The Nomex is identified as Nomex 1019 and is composed of
a mixture of polyamide and polyester. It is 4.6 mils thick, has a
weight of 3 oz/sq. yard and a Frazier air permeability of 2.5 cfm/ft2
at 1/2 inch water pressure. The elements were assembled using
Tra-bond 2125 adhesive using 9 parts resin to 1 part catalyst as per
manufacture recommendations. A diluent was added to make it less
' - 22 - 2~~7632
viscous. The diluent was Santicizer 160 plasticizer from Monsanto
which is butyl benzyl phthalate used at 10% diluent 90% Tra-bond 2125.
No surface treatment was needed for the permeate spacers although the
central tube was wiped with B.F. Goodrich A-934-BY primer to remove
any grease or dirt.
